2019年 7月 31日 (July 31) 16:30 – 18:00 @ Room 710
Lingsen Meng (UCLA)
“Application of Seismic Array Back-Projections to Rupture Imaging and Beyond”
High-frequency seismic waveforms recorded by large regional arrays have enabled back-projections (BP), an emerging tool to probe kinematic rupture processes. In this talk, I present our effort of improving the resolution and reducing the uncertainty of BP imaging, which allows us to address the open questions of earthquake source dynamics. In the case study of the 2015 Mw 8.3 Chile earthquake, we observed the splitting of rupture fronts around the rim of a large barrier. This encircling pattern is analogous to the double-pincer movement in military tactics. Such a degree of complexity is previously only seen in simulations and it is observed for the first time in real earthquakes. In the 2011 Mw 9.0 Tohoku earthquake, we find that the coseismic rupture is bounded by the bottom of the seismogenic zone, which contradicts the claims that dynamic frictional weakening enables deep penetrations of large earthquakes into the ductile creeping zone. Beyond earthquake source imaging, BP is also applied to tsunami early warning, ground-motion predictions, and monitoring of offshore seismicity. We develop a multi-array back-projection method using the local seismic network to estimate the rupture region of tsunami source. We then construct simplified source models which predict tsunami arrival times and amplitudes sufficiently accurate for warning purposes. BP’s capability of effectively capturing high-frequency seismic radiations also enable a new distance metric in the GMPEs which outperforms conventional distance metric in predicting ground shaking intensity above 1 Hz. BP can be also combined with the matched-filter approach to improve the detection of offshore seismicity. We use events retrieved by BP in the coda wave of large earthquakes as templates in the matched-filter processing to detect even more micro-seismicity. In one-year continuous data following the Tohoku earthquake, our approach finds more than 6 folds of undocumented events in the near-trench region than those listed in the JMA catalog.